Train a Tiny AI
You give the examples. The machine learns the rest. This is a real model, kept small enough to see every part of it.
Teach it two shapes
Draw an X shape or an O shape on the grid. Click cells to fill them. Then tell the model which one it is. Give it a few of each.
Fill in a shape first, then pick a label.
X examples
0
none yet
O examples
0
none yet
Give it at least one X and one O to start.
AI learns from labeled examples. You are not writing a rule like "an X has corners."
You are just showing it pictures and saying what each one is. The model figures out the pattern from your examples.
Now let it guess
Draw a new shape it has never seen. The model compares it to every example you saved and picks the closest match. Then it tells you its guess.
No magic here. The model counts how many of the 25 cells match each saved example,
keeps the one that matches best, and copies its label. That is the honest core of how a real model
is trained: learn from examples, then compare. Real ones just use far more examples and far cleverer comparing.
← AI Wing · a real model, small enough to see every moving part